hesim 0.1.0.9000

The current development version of hesim.

Highlights

hesim now provides a general framework for integrating statistical models with economic evaluation. Users can build a decision model by specifying a model structure, which consists of a set of statistical models for disease progression, utility values, and costs. Each statistical model is used to simulate outcomes as a function of estimated parameters. N-state partitioned survival models (PSMs) and individual-level continuous time state transition models (CTSTMs) are now supported.

API changes

New features

Fitted statistical models

Parameters

  • Functions prefixed by params_ create objects storing samples of parameters of fitted statistical models for probabilistic sensitivity analysis.
  • create_params() is a generic function for creating parameter objects from a fitted statistical model or a formula object. Parameters can be sampled using Monte Carlo multivariate normal approximations or via bootstrapping.
  • Current support for flexible survival modeling (params_surv(), params_surv_list()) and linear regression (params_lm()). Splines and parametric distributions (exponential, Weibull, Gompertz, gamma, lognormal log-logistic, generalized gamma) are supported for survival modeling.

Input data

  • hesim_data() creates an object of class hesim_data for storing a collection of data tables or data frames for simulation modeling.
  • expand.hesim_data() combines some or all of the data tables or data frames in hesim_data() into a single long dataset.
  • input_data() creates an object of class input_data, which contains data for predicting or simulating values with a statistical model.
  • create_input_data() creates an object of class input_data from a fitted statistical model or a formula object.

Health state values

  • The R6 class StateVals simulates the costs or utilities associated with health states.
  • create_StateVals() creates a StateVals object from fitted statistical models or formula objects.

Partitioned survival models

  • The R6 class Psm simulates outcomes from N-state PSMs.
  • A Psm object is instantiated with a set of survival models (the R6 class PsmCurves) and models for costs and utility (the R6 class StateVals).
  • create_PsmCurves() creates a PsmCurves object from fitted statistical models or formula objects.

Continuous time state transition models

  • The R6 class IndivCtstm simulates individual-level CTSTMs. Semi-Markov (i.e., “clock reset”) models are currently supported.
  • A IndivCtstm object is instantiated with a health state transition model (the R6 class CtstmTrans) and models for costs and utility (the R6 class StateVals).
  • create_CtstmTrans() creates a CtstmTrans object from fitted statistical models or formula objects.

Datasets

  • psm4_exdata provides example datasets for parameterizing a PSM.
  • ctstm3_exdata provides example datasets for parameterizing a CTSTM.

hesim 0.1.0

The initial CRAN submission.